This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
charnames: :alias alone implies :full
[perl5.git] / lib / _charnames.pm
CommitLineData
e7a078a0
KW
1# !!!!!!! INTERNAL PERL USE ONLY !!!!!!!
2# This helper module is for internal use by core Perl only. This module is
3# subject to change or removal at any time without notice. Don't use it
4# directly. Use the public <charnames> module instead.
5
6package _charnames;
7use strict;
8use warnings;
9use File::Spec;
558de9fa 10our $VERSION = '1.35';
e7a078a0
KW
11use unicore::Name; # mktables-generated algorithmically-defined names
12
13use bytes (); # for $bytes::hint_bits
14use re "/aa"; # Everything in here should be ASCII
15
16$Carp::Internal{ (__PACKAGE__) } = 1;
17
18# Translate between Unicode character names and their code points. This is a
19# submodule of package <charnames>, used to allow \N{...} to be autoloaded,
20# but it was decided not to autoload the various functions in charnames; the
21# splitting allows this behavior.
22#
23# The official names with their code points are stored in a table in
24# lib/unicore/Name.pl which is read in as a large string (almost 3/4 Mb in
25# Unicode 6.0). Each code point/name combination is separated by a \n in the
26# string. (Some of the CJK and the Hangul syllable names are determined
27# instead algorithmically via subroutines stored instead in
28# lib/unicore/Name.pm). Because of the large size of this table, it isn't
29# converted into hashes for faster lookup.
30#
31# But, user defined aliases are stored in their own hashes, as are Perl
32# extensions to the official names. These are checked first before looking at
33# the official table.
34#
35# Basically, the table is grepped for the input code point (viacode()) or
36# name (the other functions), and the corresponding value on the same line is
37# returned. The grepping is done by turning the input into a regular
38# expression. Thus, the same table does double duty, used by both name and
39# code point lookup. (If we were to have hashes, we would need two, one for
40# each lookup direction.)
41#
42# For loose name matching, the logical thing would be to have a table
43# with all the ignorable characters squeezed out, and then grep it with the
44# similiarly-squeezed input name. (And this is in fact how the lookups are
45# done with the small Perl extension hashes.) But since we need to be able to
46# go from code point to official name, the original table would still need to
47# exist. Due to the large size of the table, it was decided to not read
48# another very large string into memory for a second table. Instead, the
49# regular expression of the input name is modified to have optional spaces and
50# dashes between characters. For example, in strict matching, the regular
51# expression would be:
52# qr/\tDIGIT ONE$/m
53# Under loose matching, the blank would be squeezed out, and the re would be:
54# qr/\tD[- ]?I[- ]?G[- ]?I[- ]?T[- ]?O[- ]?N[- ]?E$/m
55# which matches a blank or dash between any characters in the official table.
56#
57# This is also how script lookup is done. Basically the re looks like
58# qr/ (?:LATIN|GREEK|CYRILLIC) (?:SMALL )?LETTER $name/
59# where $name is the loose or strict regex for the remainder of the name.
60
61# The hashes are stored as utf8 strings. This makes it easier to deal with
62# sequences. I (khw) also tried making Name.pl utf8, but it slowed things
63# down by a factor of 7. I then tried making Name.pl store the ut8
64# equivalents but not calling them utf8. That led to similar speed as leaving
65# it alone, but since that is harder for a human to parse, I left it as-is.
66
67my %system_aliases = (
e7a078a0 68
e7a078a0
KW
69 'SINGLE-SHIFT 2' => pack("U", 0x8E),
70 'SINGLE-SHIFT 3' => pack("U", 0x8F),
71 'PRIVATE USE 1' => pack("U", 0x91),
72 'PRIVATE USE 2' => pack("U", 0x92),
e7a078a0
KW
73);
74
75# These are the aliases above that differ under :loose and :full matching
76# because the :full versions have blanks or hyphens in them.
7620cb10
KW
77#my %loose_system_aliases = (
78#);
e7a078a0 79
fe3193b5
KW
80#my %deprecated_aliases;
81#$deprecated_aliases{'BELL'} = pack("U", 0x07) if $^V lt v5.17.0;
e7a078a0 82
7620cb10
KW
83#my %loose_deprecated_aliases = (
84#);
e7a078a0
KW
85
86# These are special cased in :loose matching, differing only in a medial
87# hyphen
88my $HANGUL_JUNGSEONG_O_E_utf8 = pack("U", 0x1180);
89my $HANGUL_JUNGSEONG_OE_utf8 = pack("U", 0x116C);
90
91
92my $txt; # The table of official character names
93
94my %full_names_cache; # Holds already-looked-up names, so don't have to
95# re-look them up again. The previous versions of charnames had scoping
96# bugs. For example if we use script A in one scope and find and cache
97# what Z resolves to, we can't use that cache in a different scope that
98# uses script B instead of A, as Z might be an entirely different letter
99# there; or there might be different aliases in effect in different
100# scopes, or :short may be in effect or not effect in different scopes,
101# or various combinations thereof. This was solved in this version
102# mostly by moving things to %^H. But some things couldn't be moved
103# there. One of them was the cache of runtime looked-up names, in part
104# because %^H is read-only at runtime. I (khw) don't know why the cache
105# was run-time only in the previous versions: perhaps oversight; perhaps
106# that compile time looking doesn't happen in a loop so didn't think it
107# was worthwhile; perhaps not wanting to make the cache too large. But
108# I decided to make it compile time as well; this could easily be
109# changed.
110# Anyway, this hash is not scoped, and is added to at runtime. It
111# doesn't have scoping problems because the data in it is restricted to
112# official names, which are always invariant, and we only set it and
113# look at it at during :full lookups, so is unaffected by any other
114# scoped options. I put this in to maintain parity with the older
115# version. If desired, a %short_names cache could also be made, as well
116# as one for each script, say in %script_names_cache, with each key
117# being a hash for a script named in a 'use charnames' statement. I
118# decided not to do that for now, just because it's added complication,
119# and because I'm just trying to maintain parity, not extend it.
120
121# Like %full_names_cache, but for use when :loose is in effect. There needs
122# to be two caches because :loose may not be in effect for a scope, and a
123# loose name could inappropriately be returned when only exact matching is
124# called for.
125my %loose_names_cache;
126
127# Designed so that test decimal first, and then hex. Leading zeros
128# imply non-decimal, as do non-[0-9]
129my $decimal_qr = qr/^[1-9]\d*$/;
130
131# Returns the hex number in $1.
132my $hex_qr = qr/^(?:[Uu]\+|0[xX])?([[:xdigit:]]+)$/;
133
134sub croak
135{
136 require Carp; goto &Carp::croak;
137} # croak
138
139sub carp
140{
141 require Carp; goto &Carp::carp;
142} # carp
143
144sub alias (@) # Set up a single alias
145{
225fb84f
KW
146 my @errors;
147
e7a078a0 148 my $alias = ref $_[0] ? $_[0] : { @_ };
225fb84f
KW
149 foreach my $name (sort keys %$alias) { # Sort only because it helps having
150 # deterministic output for
151 # t/lib/charnames/alias
e7a078a0
KW
152 my $value = $alias->{$name};
153 next unless defined $value; # Omit if screwed up.
154
155 # Is slightly slower to just after this statement see if it is
156 # decimal, since we already know it is after having converted from
157 # hex, but makes the code easier to maintain, and is called
158 # infrequently, only at compile-time
159 if ($value !~ $decimal_qr && $value =~ $hex_qr) {
160 $value = CORE::hex $1;
161 }
162 if ($value =~ $decimal_qr) {
013b94b6 163 no warnings qw(non_unicode surrogate nonchar); # Allow any of these
e7a078a0
KW
164 $^H{charnames_ord_aliases}{$name} = pack("U", $value);
165
166 # Use a canonical form.
167 $^H{charnames_inverse_ords}{sprintf("%05X", $value)} = $name;
168 }
169 else {
bde9e88d
KW
170 # This regex needs to be sync'd with the code in toke.c that checks
171 # for the same thing
225fb84f
KW
172 if ($name !~ / ^
173 \p{_Perl_Charname_Begin}
174 \p{_Perl_Charname_Continue}*
175 $ /x) {
176 push @errors, $name;
177 }
178 else {
179 $^H{charnames_name_aliases}{$name} = $value;
180 }
e7a078a0
KW
181 }
182 }
225fb84f
KW
183
184 # We find and output all errors from this :alias definition, rather than
185 # failing on the first one, so fewer runs are needed to get it to compile
186 if (@errors) {
187 foreach my $name (@errors) {
188 my $ok = "";
189 $ok = $1 if $name =~ / ^ ( \p{Alpha} [-\p{XPosixWord} ():\xa0]* ) /x;
190 my $first_bad = substr($name, length($ok), 1);
191 $name = "Invalid character in charnames alias definition; marked by <-- HERE in '$ok$first_bad<-- HERE " . substr($name, length($ok) + 1) . "'";
192 }
193 croak join "\n", @errors;
194 }
195
196 return;
e7a078a0
KW
197} # alias
198
199sub not_legal_use_bytes_msg {
200 my ($name, $utf8) = @_;
201 my $return;
202
203 if (length($utf8) == 1) {
204 $return = sprintf("Character 0x%04x with name '%s' is", ord $utf8, $name);
205 } else {
206 $return = sprintf("String with name '%s' (and ordinals %s) contains character(s)", $name, join(" ", map { sprintf "0x%04X", ord $_ } split(//, $utf8)));
207 }
208 return $return . " above 0xFF with 'use bytes' in effect";
209}
210
211sub alias_file ($) # Reads a file containing alias definitions
212{
213 my ($arg, $file) = @_;
214 if (-f $arg && File::Spec->file_name_is_absolute ($arg)) {
215 $file = $arg;
216 }
695481e2 217 elsif ($arg =~ m/ ^ \p{_Perl_IDStart} \p{_Perl_IDCont}* $/x) {
e7a078a0
KW
218 $file = "unicore/${arg}_alias.pl";
219 }
220 else {
013b94b6 221 croak "Charnames alias file names can only have identifier characters";
e7a078a0
KW
222 }
223 if (my @alias = do $file) {
224 @alias == 1 && !defined $alias[0] and
225 croak "$file cannot be used as alias file for charnames";
226 @alias % 2 and
227 croak "$file did not return a (valid) list of alias pairs";
228 alias (@alias);
229 return (1);
230 }
231 0;
232} # alias_file
233
234# For use when don't import anything. This structure must be kept in
235# sync with the one that import() fills up.
236my %dummy_H = (
237 charnames_stringified_names => "",
238 charnames_stringified_ords => "",
239 charnames_scripts => "",
240 charnames_full => 1,
241 charnames_loose => 0,
242 charnames_short => 0,
243 );
244
245
246sub lookup_name ($$$) {
247 my ($name, $wants_ord, $runtime) = @_;
248
249 # Lookup the name or sequence $name in the tables. If $wants_ord is false,
250 # returns the string equivalent of $name; if true, returns the ordinal value
251 # instead, but in this case $name must not be a sequence; otherwise undef is
252 # returned and a warning raised. $runtime is 0 if compiletime, otherwise
253 # gives the number of stack frames to go back to get the application caller
254 # info.
255 # If $name is not found, returns undef in runtime with no warning; and in
256 # compiletime, the Unicode replacement character, with a warning.
257
258 # It looks first in the aliases, then in the large table of official Unicode
259 # names.
260
261 my $utf8; # The string result
262 my $save_input;
263
264 if ($runtime) {
265
266 my $hints_ref = (caller($runtime))[10];
267
268 # If we didn't import anything (which happens with 'use charnames ()',
269 # substitute a dummy structure.
270 $hints_ref = \%dummy_H if ! defined $hints_ref
271 || (! defined $hints_ref->{charnames_full}
272 && ! defined $hints_ref->{charnames_loose});
273
274 # At runtime, but currently not at compile time, $^H gets
275 # stringified, so un-stringify back to the original data structures.
276 # These get thrown away by perl before the next invocation
277 # Also fill in the hash with the non-stringified data.
278 # N.B. New fields must be also added to %dummy_H
279
280 %{$^H{charnames_name_aliases}} = split ',',
281 $hints_ref->{charnames_stringified_names};
282 %{$^H{charnames_ord_aliases}} = split ',',
283 $hints_ref->{charnames_stringified_ords};
284 $^H{charnames_scripts} = $hints_ref->{charnames_scripts};
285 $^H{charnames_full} = $hints_ref->{charnames_full};
286 $^H{charnames_loose} = $hints_ref->{charnames_loose};
287 $^H{charnames_short} = $hints_ref->{charnames_short};
288 }
289
290 my $loose = $^H{charnames_loose};
291 my $lookup_name; # Input name suitably modified for grepping for in the
292 # table
293
294 # User alias should be checked first or else can't override ours, and if we
295 # were to add any, could conflict with theirs.
296 if (exists $^H{charnames_ord_aliases}{$name}) {
297 $utf8 = $^H{charnames_ord_aliases}{$name};
298 }
299 elsif (exists $^H{charnames_name_aliases}{$name}) {
300 $name = $^H{charnames_name_aliases}{$name};
301 $save_input = $lookup_name = $name; # Cache the result for any error
302 # message
303 # The aliases are documented to not match loosely, so change loose match
304 # into full.
305 if ($loose) {
306 $loose = 0;
307 $^H{charnames_full} = 1;
308 }
309 }
310 else {
311
312 # Here, not a user alias. That means that loose matching may be in
313 # effect; will have to modify the input name.
314 $lookup_name = $name;
315 if ($loose) {
316 $lookup_name = uc $lookup_name;
317
318 # Squeeze out all underscores
319 $lookup_name =~ s/_//g;
320
321 # Remove all medial hyphens
322 $lookup_name =~ s/ (?<= \S ) - (?= \S )//gx;
323
324 # Squeeze out all spaces
325 $lookup_name =~ s/\s//g;
326 }
327
328 # Here, $lookup_name has been modified as necessary for looking in the
329 # hashes. Check the system alias files next. Most of these aliases are
330 # the same for both strict and loose matching. To save space, the ones
331 # which differ are in their own separate hash, which is checked if loose
332 # matching is selected and the regular match fails. To save time, the
333 # loose hashes could be expanded to include all aliases, and there would
334 # only have to be one check. But if someone specifies :loose, they are
335 # interested in convenience over speed, and the time for this second check
336 # is miniscule compared to the rest of the routine.
337 if (exists $system_aliases{$lookup_name}) {
338 $utf8 = $system_aliases{$lookup_name};
339 }
7620cb10
KW
340 # There are currently no entries in this hash, so don't waste time looking
341 # for them. But the code is retained for the unlikely possibility that
342 # some will be added in the future.
343# elsif ($loose && exists $loose_system_aliases{$lookup_name}) {
344# $utf8 = $loose_system_aliases{$lookup_name};
345# }
fe3193b5
KW
346# if (exists $deprecated_aliases{$lookup_name}) {
347# require warnings;
348# warnings::warnif('deprecated',
349# "Unicode character name \"$name\" is deprecated, use \""
350# . viacode(ord $deprecated_aliases{$lookup_name})
351# . "\" instead");
352# $utf8 = $deprecated_aliases{$lookup_name};
353# }
7620cb10
KW
354 # There are currently no entries in this hash, so don't waste time looking
355 # for them. But the code is retained for the unlikely possibility that
356 # some will be added in the future.
357# elsif ($loose && exists $loose_deprecated_aliases{$lookup_name}) {
358# require warnings;
359# warnings::warnif('deprecated',
360# "Unicode character name \"$name\" is deprecated, use \""
361# . viacode(ord $loose_deprecated_aliases{$lookup_name})
362# . "\" instead");
363# $utf8 = $loose_deprecated_aliases{$lookup_name};
364# }
e7a078a0
KW
365 }
366
367 my @off; # Offsets into table of pattern match begin and end
368
369 # If haven't found it yet...
370 if (! defined $utf8) {
371
372 # See if has looked this input up earlier.
373 if (! $loose && $^H{charnames_full} && exists $full_names_cache{$name}) {
374 $utf8 = $full_names_cache{$name};
375 }
376 elsif ($loose && exists $loose_names_cache{$name}) {
377 $utf8 = $loose_names_cache{$name};
378 }
379 else { # Here, must do a look-up
380
381 # If full or loose matching succeeded, points to where to cache the
382 # result
383 my $cache_ref;
384
385 ## Suck in the code/name list as a big string.
386 ## Lines look like:
387 ## "00052\tLATIN CAPITAL LETTER R\n"
388 # or
389 # "0052 0303\tLATIN CAPITAL LETTER R WITH TILDE\n"
390 $txt = do "unicore/Name.pl" unless $txt;
391
392 ## @off will hold the index into the code/name string of the start and
393 ## end of the name as we find it.
394
395 ## If :loose, look for a loose match; if :full, look for the name
396 ## exactly
397 # First, see if the name is one which is algorithmically determinable.
398 # The subroutine is included in Name.pl. The table contained in
399 # $txt doesn't contain these. Experiments show that checking
400 # for these before checking for the regular names has no
401 # noticeable impact on performance for the regular names, but
402 # the other way around slows down finding these immensely.
403 # Algorithmically determinables are not placed in the cache because
404 # that uses up memory, and finding these again is fast.
405 if (($loose || $^H{charnames_full})
406 && (defined (my $ord = charnames::name_to_code_point_special($lookup_name, $loose))))
407 {
408 $utf8 = pack("U", $ord);
409 }
410 else {
411
412 # Not algorithmically determinable; look up in the table. The name
413 # will be turned into a regex, so quote any meta characters.
414 $lookup_name = quotemeta $lookup_name;
415
416 if ($loose) {
417
418 # For loose matches, $lookup_name has already squeezed out the
419 # non-essential characters. We have to add in code to make the
420 # squeezed version match the non-squeezed equivalent in the table.
421 # The only remaining hyphens are ones that start or end a word in
422 # the original. They have been quoted in $lookup_name so they look
423 # like "\-". Change all other characters except the backslash
424 # quotes for any metacharacters, and the final character, so that
425 # e.g., COLON gets transformed into: /C[- ]?O[- ]?L[- ]?O[- ]?N/
426 $lookup_name =~ s/ (?! \\ -) # Don't do this to the \- sequence
427 ( [^-\\] ) # Nor the "-" within that sequence,
428 # nor the "\" that quotes metachars,
429 # but otherwise put the char into $1
430 (?=.) # And don't do it for the final char
431 /$1\[- \]?/gx; # And add an optional blank or
432 # '-' after each $1 char
433
434 # Those remaining hyphens were originally at the beginning or end of
435 # a word, so they can match either a blank before or after, but not
436 # both. (Keep in mind that they have been quoted, so are a '\-'
437 # sequence)
438 $lookup_name =~ s/\\ -/(?:- | -)/xg;
439 }
440
441 # Do the lookup in the full table if asked for, and if succeeds
442 # save the offsets and set where to cache the result.
443 if (($loose || $^H{charnames_full}) && $txt =~ /\t$lookup_name$/m) {
444 @off = ($-[0] + 1, $+[0]); # The 1 is for the tab
445 $cache_ref = ($loose) ? \%loose_names_cache : \%full_names_cache;
446 }
447 else {
448
449 # Here, didn't look for, or didn't find the name.
450 # If :short is allowed, see if input is like "greek:Sigma".
451 # Keep in mind that $lookup_name has had the metas quoted.
452 my $scripts_trie = "";
453 my $name_has_uppercase;
454 if (($^H{charnames_short})
455 && $lookup_name =~ /^ (?: \\ \s)* # Quoted space
456 (.+?) # $1 = the script
457 (?: \\ \s)*
458 \\ : # Quoted colon
459 (?: \\ \s)*
460 (.+?) # $2 = the name
461 (?: \\ \s)* $
462 /xs)
463 {
464 # Even in non-loose matching, the script traditionally has been
465 # case insensitve
466 $scripts_trie = "\U$1";
467 $lookup_name = $2;
468
469 # Use original name to find its input casing, but ignore the
470 # script part of that to make the determination.
471 $save_input = $name if ! defined $save_input;
472 $name =~ s/.*?://;
473 $name_has_uppercase = $name =~ /[[:upper:]]/;
474 }
475 else { # Otherwise look in allowed scripts
476 $scripts_trie = $^H{charnames_scripts};
477
478 # Use original name to find its input casing
479 $name_has_uppercase = $name =~ /[[:upper:]]/;
480 }
481
482 my $case = $name_has_uppercase ? "CAPITAL" : "SMALL";
90249f0a
KW
483 return if (! $scripts_trie || $txt !~
484 /\t (?: $scripts_trie ) \ (?:$case\ )? LETTER \ \U$lookup_name $/xm);
e7a078a0
KW
485
486 # Here have found the input name in the table.
487 @off = ($-[0] + 1, $+[0]); # The 1 is for the tab
488 }
489
490 # Here, the input name has been found; we haven't set up the output,
491 # but we know where in the string
492 # the name starts. The string is set up so that for single characters
493 # (and not named sequences), the name is preceded immediately by a
494 # tab and 5 hex digits for its code, with a \n before those. Named
495 # sequences won't have the 7th preceding character be a \n.
496 # (Actually, for the very first entry in the table this isn't strictly
497 # true: subtracting 7 will yield -1, and the substr below will
498 # therefore yield the very last character in the table, which should
499 # also be a \n, so the statement works anyway.)
500 if (substr($txt, $off[0] - 7, 1) eq "\n") {
501 $utf8 = pack("U", CORE::hex substr($txt, $off[0] - 6, 5));
502
503 # Handle the single loose matching special case, in which two names
504 # differ only by a single medial hyphen. If the original had a
505 # hyphen (or more) in the right place, then it is that one.
506 $utf8 = $HANGUL_JUNGSEONG_O_E_utf8
507 if $loose
508 && $utf8 eq $HANGUL_JUNGSEONG_OE_utf8
509 && $name =~ m/O \s* - [-\s]* E/ix;
510 # Note that this wouldn't work if there were a 2nd
511 # OE in the name
512 }
513 else {
514
515 # Here, is a named sequence. Need to go looking for the beginning,
516 # which is just after the \n from the previous entry in the table.
517 # The +1 skips past that newline, or, if the rindex() fails, to put
518 # us to an offset of zero.
519 my $charstart = rindex($txt, "\n", $off[0] - 7) + 1;
520 $utf8 = pack("U*", map { CORE::hex }
521 split " ", substr($txt, $charstart, $off[0] - $charstart - 1));
522 }
523 }
524
525 # Cache the input so as to not have to search the large table
526 # again, but only if it came from the one search that we cache.
527 # (Haven't bothered with the pain of sorting out scoping issues for the
528 # scripts searches.)
529 $cache_ref->{$name} = $utf8 if defined $cache_ref;
530 }
531 }
532
533
534 # Here, have the utf8. If the return is to be an ord, must be any single
535 # character.
536 if ($wants_ord) {
537 return ord($utf8) if length $utf8 == 1;
538 }
539 else {
540
541 # Here, wants string output. If utf8 is acceptable, just return what
542 # we've got; otherwise attempt to convert it to non-utf8 and return that.
543 my $in_bytes = ($runtime)
544 ? (caller $runtime)[8] & $bytes::hint_bits
545 : $^H & $bytes::hint_bits;
546 return $utf8 if (! $in_bytes || utf8::downgrade($utf8, 1)) # The 1 arg
547 # means don't die on failure
548 }
549
550 # Here, there is an error: either there are too many characters, or the
551 # result string needs to be non-utf8, and at least one character requires
552 # utf8. Prefer any official name over the input one for the error message.
553 if (@off) {
554 $name = substr($txt, $off[0], $off[1] - $off[0]) if @off;
555 }
556 else {
557 $name = (defined $save_input) ? $save_input : $_[0];
558 }
559
560 if ($wants_ord) {
561 # Only way to get here in this case is if result too long. Message
562 # assumes that our only caller that requires single char result is
563 # vianame.
564 carp "charnames::vianame() doesn't handle named sequences ($name). Use charnames::string_vianame() instead";
565 return;
566 }
567
568 # Only other possible failure here is from use bytes.
569 if ($runtime) {
570 carp not_legal_use_bytes_msg($name, $utf8);
571 return;
572 } else {
573 croak not_legal_use_bytes_msg($name, $utf8);
574 }
575
576} # lookup_name
577
578sub charnames {
579
580 # For \N{...}. Looks up the character name and returns the string
581 # representation of it.
582
583 # The first 0 arg means wants a string returned; the second that we are in
584 # compile time
585 return lookup_name($_[0], 0, 0);
586}
587
588sub import
589{
590 shift; ## ignore class name
591
592 if (not @_) {
593 carp("'use charnames' needs explicit imports list");
594 }
595 $^H{charnames} = \&charnames ;
596 $^H{charnames_ord_aliases} = {};
597 $^H{charnames_name_aliases} = {};
598 $^H{charnames_inverse_ords} = {};
599 # New fields must be added to %dummy_H, and the code in lookup_name()
600 # that copies fields from the runtime structure
601
602 ##
603 ## fill %h keys with our @_ args.
604 ##
605 my ($promote, %h, @args) = (0);
606 while (my $arg = shift) {
607 if ($arg eq ":alias") {
608 @_ or
609 croak ":alias needs an argument in charnames";
610 my $alias = shift;
611 if (ref $alias) {
612 ref $alias eq "HASH" or
613 croak "Only HASH reference supported as argument to :alias";
614 alias ($alias);
5198c137 615 $promote = 1;
e7a078a0
KW
616 next;
617 }
618 if ($alias =~ m{:(\w+)$}) {
619 $1 eq "full" || $1 eq "loose" || $1 eq "short" and
620 croak ":alias cannot use existing pragma :$1 (reversed order?)";
621 alias_file ($1) and $promote = 1;
622 next;
623 }
5198c137 624 alias_file ($alias) and $promote = 1;
e7a078a0
KW
625 next;
626 }
627 if (substr($arg, 0, 1) eq ':'
628 and ! ($arg eq ":full" || $arg eq ":short" || $arg eq ":loose"))
629 {
630 warn "unsupported special '$arg' in charnames";
631 next;
632 }
633 push @args, $arg;
634 }
635
636 @args == 0 && $promote and @args = (":full");
637 @h{@args} = (1) x @args;
638
639 # Don't leave these undefined as are tested for in lookup_names
640 $^H{charnames_full} = delete $h{':full'} || 0;
641 $^H{charnames_loose} = delete $h{':loose'} || 0;
642 $^H{charnames_short} = delete $h{':short'} || 0;
643 my @scripts = map { uc quotemeta } keys %h;
644
645 ##
646 ## If utf8? warnings are enabled, and some scripts were given,
647 ## see if at least we can find one letter from each script.
648 ##
649 if (warnings::enabled('utf8') && @scripts) {
650 $txt = do "unicore/Name.pl" unless $txt;
651
652 for my $script (@scripts) {
653 if (not $txt =~ m/\t$script (?:CAPITAL |SMALL )?LETTER /) {
654 warnings::warn('utf8', "No such script: '$script'");
655 $script = quotemeta $script; # Escape it, for use in the re.
656 }
657 }
658 }
659
660 # %^H gets stringified, so serialize it ourselves so can extract the
661 # real data back later.
662 $^H{charnames_stringified_ords} = join ",", %{$^H{charnames_ord_aliases}};
663 $^H{charnames_stringified_names} = join ",", %{$^H{charnames_name_aliases}};
664 $^H{charnames_stringified_inverse_ords} = join ",", %{$^H{charnames_inverse_ords}};
665
666 # Modify the input script names for loose name matching if that is also
667 # specified, similar to the way the base character name is prepared. They
668 # don't (currently, and hopefully never will) have dashes. These go into a
669 # regex, and have already been uppercased and quotemeta'd. Squeeze out all
670 # input underscores, blanks, and dashes. Then convert so will match a blank
671 # between any characters.
672 if ($^H{charnames_loose}) {
673 for (my $i = 0; $i < @scripts; $i++) {
674 $scripts[$i] =~ s/[_ -]//g;
675 $scripts[$i] =~ s/ ( [^\\] ) (?= . ) /$1\\ ?/gx;
676 }
677 }
678
679 $^H{charnames_scripts} = join "|", @scripts; # Stringifiy them as a trie
680} # import
681
682# Cache of already looked-up values. This is set to only contain
683# official values, and user aliases can't override them, so scoping is
684# not an issue.
685my %viacode;
686
687sub viacode {
688
689 # Returns the name of the code point argument
690
691 if (@_ != 1) {
692 carp "charnames::viacode() expects one argument";
693 return;
694 }
695
696 my $arg = shift;
697
698 # This is derived from Unicode::UCD, where it is nearly the same as the
699 # function _getcode(), but here it makes sure that even a hex argument
700 # has the proper number of leading zeros, which is critical in
701 # matching against $txt below
702 # Must check if decimal first; see comments at that definition
703 my $hex;
704 if ($arg =~ $decimal_qr) {
705 $hex = sprintf "%05X", $arg;
706 } elsif ($arg =~ $hex_qr) {
707 # Below is the line that differs from the _getcode() source
708 $hex = sprintf "%05X", hex $1;
709 } else {
710 carp("unexpected arg \"$arg\" to charnames::viacode()");
711 return;
712 }
713
714 return $viacode{$hex} if exists $viacode{$hex};
715
7620cb10
KW
716 my $return;
717
e7a078a0
KW
718 # If the code point is above the max in the table, there's no point
719 # looking through it. Checking the length first is slightly faster
720 if (length($hex) <= 5 || CORE::hex($hex) <= 0x10FFFF) {
721 $txt = do "unicore/Name.pl" unless $txt;
722
723 # See if the name is algorithmically determinable.
724 my $algorithmic = charnames::code_point_to_name_special(CORE::hex $hex);
725 if (defined $algorithmic) {
726 $viacode{$hex} = $algorithmic;
727 return $algorithmic;
728 }
729
730 # Return the official name, if exists. It's unclear to me (khw) at
731 # this juncture if it is better to return a user-defined override, so
732 # leaving it as is for now.
733 if ($txt =~ m/^$hex\t/m) {
734
735 # The name starts with the next character and goes up to the
736 # next new-line. Using capturing parentheses above instead of
737 # @+ more than doubles the execution time in Perl 5.13
7620cb10
KW
738 $return = substr($txt, $+[0], index($txt, "\n", $+[0]) - $+[0]);
739
740 # If not one of these 4 code points, return what we've found.
741 if ($hex !~ / ^ 000 (?: 8[014] | 99 ) $ /x) {
742 $viacode{$hex} = $return;
743 return $return;
744 }
745
746 # For backwards compatibility, we don't return the official name of
747 # the 4 code points if there are user-defined aliases for them -- so
748 # continue looking.
e7a078a0
KW
749 }
750 }
751
752 # See if there is a user name for it, before giving up completely.
753 # First get the scoped aliases, give up if have none.
754 my $H_ref = (caller(1))[10];
7620cb10
KW
755 return if ! defined $return
756 && (! defined $H_ref
757 || ! exists $H_ref->{charnames_stringified_inverse_ords});
e7a078a0 758
424313d4
KW
759 my %code_point_aliases;
760 if (defined $H_ref->{charnames_stringified_inverse_ords}) {
761 %code_point_aliases = split ',',
e7a078a0 762 $H_ref->{charnames_stringified_inverse_ords};
424313d4
KW
763 return $code_point_aliases{$hex} if exists $code_point_aliases{$hex};
764 }
7620cb10 765
de72f72f
KW
766 # Here there is no user-defined alias, return any official one.
767 return $return if defined $return;
7620cb10 768
83adc448
KW
769 if (CORE::hex($hex) > 0x10FFFF
770 && warnings::enabled('non_unicode'))
771 {
de72f72f
KW
772 carp "Unicode characters only allocated up to U+10FFFF (you asked for U+$hex)";
773 }
774 return;
e7a078a0 775
e7a078a0
KW
776} # _viacode
777
7781;
779
780# ex: set ts=8 sts=2 sw=2 et: